Rank-biserial correlation for two independent samples

The rank-biserial correlation derived from the Mann-Whitney U statistic, equal to 2*U1/(n1*n2) - 1 and to 1 - 2*U2/(n1*n2) (Kerby 2014), where U1 counts the pairs in which the first sample exceeds the second. For two independent samples this equals Cliff's delta; the confidence interval is Cliff's (1993) consistent-variance interval.

Usage

rankBiserial(x, y, conf_level = 0.95)

Arguments

x, y: Numeric vectors for the two samples.
conf_level: Confidence level (default 0.95).

Value

A list with r (rank-biserial correlation), ci_lower, ci_upper, the Mann-Whitney u for x, and the sample sizes.

References

Kerby, D.S. (2014). Compr Psychol 3:11.IT.3.1. Cliff, N. (1993). Psychol Bull 114(3):494-509.

Examples

rankBiserial(c(5, 6, 7, 8), c(1, 2, 3, 9))
#> $r
#> [1] 0.5
#> 
#> $ci_lower
#> [1] -0.4878146
#> 
#> $ci_upper
#> [1] 0.9263176
#> 
#> $u
#> [1] 12
#> 
#> $n1
#> [1] 4
#> 
#> $n2
#> [1] 4
#> 
#> $conf_level
#> [1] 0.95
#>